The present invention relates to a remote conference system, and more particularly to a remote conference system which transmits voice data using multicast in a server client model.
In a conventional remote conference system, each of conference terminals and a conference controller are mutually connected. Each conference terminal transmits voice data to the conference controller. The conference controller synthesizes the voice data transmitted from each of the conference terminals, and transmits the synthesized voice data to each of the conference terminals. At the time of synthesizing the voice data, the conference controller performs echo canceling processing in order to prevent echo (e.g., howling) in each of the conference terminals.
The echo canceling processing is the one in which the voice data already transmitted from a certain conference terminal is canceled from voice data to be transmitted to that conference terminal. For example, assuming that voice data from a conference terminal A be "a", voice data from a conference terminal B be "b", and voice data from a conference terminal C be "c", voice data (b+c) is transmitted to the conference terminal A, voice data (c+a) is transmitted to the conference terminal B, and voice data (a+b) is transmitted to the conference terminal C.
In the conventional remote conference system, since the different voice data are transmitted from the conference controller to the conference terminals, the conference controller must synthesize the voice data for each conference terminal one by one. In addition, the conference controller must transmit each of the voice data one by one (i.e., unicast transmission). Therefore, loads for processing in the conference controller increase and traffics on networks increase, resulting in an increase in data delay.